Farmed Atlantic salmon (Salmo salar L.) have traditionally been fed diets containing major amounts of fish oil and fish meal. The steady increase in aquaculture production volume of 8–10 % per year(Reference Tacon, Hasan and Subasinghe1) has resulted in an increased use of alternative proteins and oils in aqua feeds. Vegetable oils are recognised as suitable alternatives to fish oils(Reference Torstensen, Bell and Rosenlund2, Reference Turchini, Torstensen and Ng3). Vegetable oils are, however, devoid of marine n-3 PUFA (EPA (20 : 5n-3), docosapentaenoic acid (22 : 5n-3) and DHA (22 : 6n-3)) while the levels of linoleic acid (18 : 2n-6) and monoene fatty acids are usually high, resulting in low dietary n-3:n-6 ratios. Likewise, sustainable alternatives to fish meal are included among plant protein sources, such as vegetable meals, with protein contents of 20 to 50 %(Reference Hertrampf and Piedad-Pascual4). However, the indispensable amino acid profile of plant proteins differs from that of fish meal. Sensible blending of different protein sources are adopted to complete the indispensable amino acid composition, while low levels of selected crystalline amino acids may also be added to fulfil amino acid requirements.
Studies on metabolic effects when replacing marine ingredients have concentrated either on fish oil(Reference Turchini, Torstensen and Ng3–Reference Leaver, Bautista and Bjornsson5) or fish meal replacement(Reference Dias, Alvarez and Arzel6–Reference Messina, Tulli and Messina8) independently. Typically, replacing fish oil with vegetable oils and thus decreasing marine n-3 PUFA and the n-3:n-6 ratio as well as increasing oleic acid (18 : 1n-9) has been reported to increase hepatic lipid stores in Atlantic salmon(Reference Jordal, Lie and Torstensen9), particularly at low water temperatures(Reference Ruyter, Moya-Falcón and Rosenlund10). Plasma cholesterol levels have been reported to decrease, whereas plasma TAG was unaffected by fish oil replacement(Reference Jordal, Lie and Torstensen9). Dietary fatty acid composition may also regulate the expression of genes encoding for lipid metabolism-related genes(Reference Leaver, Bautista and Bjornsson5–Reference Torstensen, Nanton and Olsvik11).
Some of the most frequently used vegetable oils contain high levels of 18 : 1n-9 which has been shown in different cell-culture systems(Reference Ranheim, Gedde-Dahl and Rustan12, Reference Vegusdal, Gjøen and Berge13) and rats(Reference Halvorsen, Rustan and Madsen14) to affect liver lipid and lipoprotein metabolism. EPA and DHA are reported to affect liver TAG metabolism and β-oxidation capacity(Reference Berge, Madsen and Vaagenes15–Reference Willumsen, Vaagenes and Asiedu18), with especially EPA having a plasma lipid-lowering effect in rats(Reference Frøyland, Madsen and Vaagenes19, Reference Frøyland, Vaagenes and Asiedu20). Furthermore, n-3 PUFA is suggested to inhibit the secretion of TAG-rich VLDL particles by inhibiting the rate-limiting enzyme diacylglycerol acyltransferase (EC 2.3.1.20)(Reference Berge, Madsen and Vaagenes15, Reference Madsen, Rustan and Vaagenes16) and by inhibiting the assembly of VLDL particles in the liver(Reference Lang and Davis21–Reference Kendrick and Higgins23). Several studies in human subjects have shown that dietary EPA and DHA decrease plasma TAG(Reference Harris, Connor and McMurphy24, Reference Nestel25) and protect against CHD(Reference Bang, Dyerberg and Nielsen26, Reference Seierstad, Seljeflot and Johansen27).
A large body of literature presents studies on alternative protein sources to fish meal in feed to salmonids, both by using blends of plant proteins with or without amino acid supplementation, as well as total replacement of fish meal(Reference Dias, Alvarez and Arzel6, Reference de Francesco, Parisi and Medale28–Reference Kaushik, Cravedi and Lalles32). A high inclusion of dietary plant protein mix has been reported to increase lipid retention in rainbow trout(Reference Kaushik, Covès and Dutto33) whereas the opposite effects was reported in Atlantic salmon fed diets without any fish meal(Reference Espe, Lemme and Petri29). The reduced lipid deposition disappeared when only 5 % fish meal was added to the diets(Reference Espe, Lemme and Petri30). Furthermore, suboptimal dietary levels of methionine in high-plant protein diets have been reported to increase liver TAG concentration, fatty acid synthase activity and the 18 : 1:18 : 0 ratio in Atlantic salmon(Reference Espe, Rathore and Du34), resembling the signs observed in early stages of non-alcoholic fatty liver disease in rodent models. Suboptimal lysine in high-plant protein diets increased the whole-body lipid concentration(Reference Espe, Lemme and Petri30) and affected the partitioning of growth in on-growing Atlantic salmon post-smolt(Reference Rathore, Liaset and Hevrøy35).
Replacement of both fish oil and fish meal in diets fed to Atlantic salmon has been reported to result in decreased growth, which could only partially be explained by reduced feed intake during the first 3 months(Reference Torstensen, Espe and Sanden36). The aim of the present study, which was part of a larger project(Reference Torstensen, Espe and Sanden36), was to further elucidate the metabolic effects of the combined and maximised replacement of fish meal and fish oil with vegetable feed ingredients with special emphasis on lipid metabolism.
Materials and methods
Feeding trial
The experiments complied with the guidelines of the Norwegian Regulation on Animal Experimentation and European Community Directive 86/609/EEC, and the protocol was approved by competent individuals at the laboratory unit at the Institute of Marine Research (Bergen, Norway) and the National Animal Research Authority.
The feeding trial was carried out at the Institute of Marine Research, Matre (Matredal, Norway; 60°52′N, 05°35′E) during the period from 22 June 2006 to 15 June 2007. The Atlantic salmon were obtained from AkvaGen A/S (Tingvoll, Norway). In June 2006, approximately 6000 smolt with a mean weight of 355 (sd 92) g were distributed equally into twelve 10 m3 indoor fibreglass tanks containing 7 m3 seawater, with a continuous flow-through (about 52 litres/min) of seawater (salinity 34·9 g/l) from a deepwater inlet (80 m deep, Matrefjorden). Temperature was kept constant at 8·9 ± 0·1°C, with continuous recording and an automatic record system. O2 was also automatically recorded in the outlet water and was never decreased below 80 % saturation. The fish were acclimatised to the experimental conditions for 2·5 weeks before being fed the experimental diets on 22 June 2006. Fish in three randomised tanks were fed four different extruded diets: (1) a diet with maximum inclusion of fish meal (FM) and fish oil (FMFO); (2) a diet with an estimated safe maximum replacement of both fish meal and fish oil with plant meal (80 % plant protein) and vegetable oil (70 % vegetable oil) (80PP70VO); (3) a diet with half the maximum replacement with plant meal and maximum replacement with vegetable oil (40PP70VO); and (4) a diet with maximum replacement with plant meal and half the maximum replacement with vegetable oil (80PP35VO). Diets were produced by Skretting ARC (Stavanger, Norway). The four experimental diets were fed throughout the entire seawater production phase, with all diets changing in pellet size and lipid content after a 3-month feeding period (Table 1). Diet with pellet size 4 mm was fed to the fish from 20 June 2006 to 20 September 2006, and from 20 September 2006 to 12 June 2007 the fish were fed the 6 mm diets. Capelin oil (Fish Oil Nordic, Nordsildmel, Norway) was used as the dietary fish oil, and was the main source for very-long-chain highly unsaturated n-3 PUFA. A mixture of rapeseed oil, palm oil and linseed oil (55:30:15, by vol.) was used as the replacement for fish oil (Table 1). The mixture was selected to obtain a lipid profile of SFA, MUFA and n-3 PUFA as close as possible to capelin oil(Reference Torstensen, Bell and Rosenlund2).
Table 1 Feed composition (g/kg) and proximate composition (fat, protein, starch, ash and DM (g 100 g/g wet weight) and energy as kJ/g) of two pellet sizes of the four experimental diets
FMFO, 100 % fish meal and 100 % fish oil; 80PP35VO, 80 % plant protein and 35 % vegetable oil blend; 40PP70VO, 40 % plant protein and 70 % vegetable oil blend; 80PP70VO, 80 % plant protein and 70 % vegetable oil blend; LT, low temperature .
* Wheat: Statkorn, Oslo, Norway; wheat gluten: Cerestar Scandinavia AS, Copenhagen, Denmark; maize gluten: Cargill, Minneapolis, MN, USA; soyabean meal extracted: Denofa, Fredrikstad, Norway; krill meal: Aker Seafoods ASA, Oslo, Norway; LT South American: Consortio, Arequipa, Peru; linseed oil: Elbe Fetthandel GmbH, Geesthacht, Germany; palm oil: Denofa; rapeseed oil: Emmelev AS, Otterup, Denmark; fish oil: Nordsildmel, Fyllingsdalen, Norway.
† Vitamin and mineral supplementation is estimated to meet the requirements according to National Research Council(39) recommendations.
‡ The method uncertainty is based on internal reproducibility, analyses of reference material and between-laboratory test of quality-assured methods.
As replacement for fish meal, a mixture of maize gluten, wheat gluten, soya concentrate and krill meal was used (Table 1). A small component ( < 50 g/kg) of krill meal (Aker Seafoods ASA, Oslo, Norway) was added to the experimental diets, with the aim to improve palatability and thus voluntary feed intake(Reference Gaber37, Reference Olsen, Suontama and Langmyhr38). All diets were formulated to meet nutrient requirements of fish according to the National Research Council(39) recommendations. After 0, 3, 5, 8 and 12 months fish were weighed, and the amount of feed given was adjusted according to their biomass. Fish were reared under a natural light regimen until October 2006 when a 10 h light–14 h dark regimen was maintained throughout the winter until March 2007, when the fish returned to the natural light regimen. From June 2006 until February 2007, the feed consumption per tank was recorded daily. Fish were fed in excess twice per d (at 07.00 and 14.00 hours) with automatic feeders for 30 min, followed by feed collection 30 min after each feeding. Measuring of individual weights of at least 30 % of the biomass per tank and sampling of fish for whole-body analysis (six pooled fish per tank) were performed in June 2006 (initial stage), September 2006 (T = 3 months), November 2006 (T = 5 months), February 2007 (T = 8 months) and June 2007 (T = 12 months). Measuring individual weights of all fish at each sampling point was avoided to minimise handling stress and ensure optimal growth conditions. Growth, nutrient retention and digestibility are reported elsewhere(Reference Torstensen, Espe and Sanden36).
Sampling
Samples of each experimental diet and the oil and meal ingredients used in the feeds for each feed production batch were stored at − 20°C. From each tank, eleven fish were anaesthetised with benzocaine (7 g/l) and killed by a blow to the head. Of these, three fish from each tank were sampled for whole-body proximate composition, plasma lipids and lipoprotein separations, and these were not fed 2 d before sampling. Blood was collected from the caudal vein using EDTA vacutainers for plasma samples for further lipoprotein separation and heparin for the plasma samples for further analyses of nutrients. At 6 h after the last feeding, samples were collected from five fish per tank and pooled for tissue lipid and amino acid analyses and the determination of plasma amino acids. Visceral fat index was based on the visceral weight of eight fish from each tank (totally twenty-four fish from each dietary treatment). For RNA isolation and gene expression analyses, liver and visceral adipose tissue from three fish from each tank (in total nine fish per dietary treatment and collected 6 h after last feeding) were sampled individually and flash-frozen on liquid N2. All samples were stored at − 80°C until analysed.
Dietary amino acids and nitrogen metabolites in liver
Dietary amino acids were analysed after hydrolysis in 6 m-HCl as described previously for the growth and accretion data of the present trial(Reference Torstensen, Espe and Sanden36). N metabolites in the liver at the end of the feeding experiment were extracted and analysed as described(Reference Espe, Lemme and Petri29).
Lipid extraction and fatty acid analysis
The lipid content of diets, whole fish and flesh was determined gravimetrically as the sum of free and bound fat. Free or loosely bound fat was extracted with petroleum ether and dried at 103 ± 1°C. The samples were thereafter hydrolysed with HCl in a Tecator Soxtec Hydrolysing unit (Foss Tecator AB, Höganös, Sweden) to release the bound fat, which was extracted with petroleum ether and dried at 103 ± 1°C. Total lipid was extracted from diets, liver and flesh by homogenisation in chloroform–methanol (2:1, v/v) with 19 : 0 methyl ester as the internal standard. Fatty acid methyl esters were prepared from total lipid by boron trifluoride following saponification, as described previously(Reference Lie and Lambertsen40, Reference Torstensen, Frøyland and Lie41). Thermo Finnegan Trace 2000 GC equipped with a fused silica capillary column was used (CP-sil 88; 50 m × 0·32 mm internal diameter; Chrompak Ltd, Middelburg, The Netherlands) with temperature programming of 60°C for 1 min, 160°C for 28 min, 190°C for 17 min, and finally 220°C for 10 min with all intervening temperature ramps being at 25°C per min. Individual methyl esters were identified by comparison with known standards and on the basis of published values(Reference Ackman42). Data were collected and processed using Totalchrom software (version 6.2; PerkinElmer, Waltham, MA, USA).
Lipid classes of salmon liver and diets were determined essentially as described by Jordal et al. (Reference Jordal, Lie and Torstensen9) based on Bell et al. (Reference Bell, Dick and McVicar43). Briefly, lipids were extracted from the liver and diets by homogenising in chloroform–methanol (2:1, v/v) with 0·01 % butylated hydroxytoluene (BHT). The samples were filtrated, evaporated after the addition of isopropanol and dissolved in chloroform–0·01 % BHT before separation on high-performance TLC plates. Total lipid (10 μg) was applied to a 10 × 20 cm high-performance TLC plate that had been pre-run in hexane–diethyl ether (1:1, v/v) and activated at 110°C for 30 min. The plates were developed at 5·5 cm in methyl acetate–isopropanol–chloroform–methanol–0·25 % (w/v) aqueous KCl (25:25:25:10:9, by vol.) to separate phospholipid classes with neutral lipids running at the solvent front(Reference Vitello and Zanetta44). After drying, the plates were developed fully in hexane–diethyl ether–acetic acid (80:20:2, by vol.) to separate neutral lipids and cholesterol. Lipid classes were visualised by charring at 160°C for 15 min after spraying with 3 % copper acetate (w/v) in 8 % (v/v) phosphoric acid and identified by comparison with commercially available standards. Lipid classes were quantified by scanning densitometry using a CAMAG TLC Scanner 3 and calculated using an integrator (WinCATS-Planar Chromatography, version 1.2.0; CAMAG, Berlin, Germany). Quantitative determination (mg lipid class/g tissue) of lipid classes was performed by establishing standard equations for each lipid class within a linear area, in addition to including a standard mixture of all the lipid classes at each high-performance TLC plate for corrections between plate variations.
Plasma and lipoprotein lipids were analysed using a clinical bioanalyser (Maxmat PL analyser; MaxMat S.A., Montpellier, France) according to standardised procedures, reagents and controls. Plasma lipoproteins (VLDL, LDL and HDL) in plasma were obtained by sequential centrifugal flotation(Reference Havel, Eder and Havel45, Reference Aviram46) as described by Lie et al. (Reference Lie, Sandvin and Waagboe47) using a Beckman Optima™XL-100K Ultracentrifuge equipped with a SW41Ti rotor. The centrifugation was done at 197 600 g av and 4°C. The density intervals were obtained by the addition of solid KBr(Reference Warnick, Cheung and Albers48), and the run times for the separation of lipoproteins were: VLDL, density (d) < 1·015 g/ml for 20 h; LDL, 1·015 g/ml < d < 1·085 g/ml for 20 h; and HDL, 1·085 g/ml < d < 1·21 g/ml for 44 h. The lipoprotein fractions were stored at − 80°C until further analysis.
For gene expression analysis total RNA was isolated from 100–500 μg of tissue sample by the standard TRIzol extraction method (Invitrogen Ltd, Paisley, Renfrewshire, UK) and recovered in 100 μl molecular diethylpyrocarbonate (DEPC)-treated water. In order to remove any possible genomic DNA contamination, the total RNA samples were pretreated using DNA-free™ DNase treatment and removal reagents kit (Ambion Inc., Austin, TX, USA) following the manufacturer's protocol.
A two-step real-time RT-PCR protocol was developed to measure the mRNA levels of the target genes in Atlantic salmon liver and visceral adipose tissue. The RT reactions were run in triplicates on ninety-six-well reaction plates with the GeneAmp PCR 9700 instrument (Applied Biosystems, Foster City, CA, USA) using TaqMan Reverse Transcription Reagent containing Multiscribe RT (50 U/μl) (N808-0234; Applied Biosystems). For efficiency calculations, twofold serial dilutions of total RNA were made. A dilution curve was recorded using four serial dilutions (250–31 ng), with each concentration being run in triplicate. The samples were analysed by quantitative RT-PCR in separate sample wells and the resulting cycle thresholds (Ct) recorded. Total RNA input was 125 ng in each reaction for all genes. No template control (ntc) and RT-control (a duplicate RNA sample analysis where only the RT enzyme is left out) reactions were run for quality assessment. RT-controls were not performed for each sample, but were run for each assay or gene, with the same sample as used to make the dilution curves on the ninety-six-well plates. Reverse transcription was performed at 48°C for 60 min by using oligo dT primers (2·5 μm) in 50 μl total volume. The final concentrations of the other reagents in the RT reaction were: MgCl2 (5·5 mm), dNTP (500 μm of each), 10X TaqMan RT buffer (1X), RNase inhibitor (0·4 U/μl) and Multiscribe RT (1·67 U/μl). cDNA (2·0 μl) from each RT reaction for all genes was transferred to a new ninety-six-well reaction plate, and the real-time PCR run in 20 μl reactions on the LightCycler® 480 Real-Time PCR System (Roche Applied Sciences, Basel, Switzerland). The final concentration of the primers was 500 nm. Real-time PCR was performed using TaqMan universal PCR master mix (LightCycler 480 SYBR Green master mix kit; Roche Applied Sciences) containing FastStart DNA polymerase, and gene-specific primers. PCR was achieved with a 5 min activation and denaturising step at 95°C, followed by forty cycles of a 15 s denaturing step at 95°C, a 60 s annealing step and a 30 s synthesis step at 72°C. The annealing temperature and sequences for the primer pairs are presented in Table 2.
Table 2 Primers for quantitative PCR assays*
FATP1, fatty acid transport protein 1; LPL, lipoprotein lipase; FABP, fatty acid binding protein; ARP, acidic ribosomal phosphoprotein; EF1AB, elongation factor 1Aβ.
* All sequences are presented as 5′ to 3′.
Data analyses and statistics
Q-Gene was used for the normalisation and calculation of relative expression data(Reference Simon49). Q-Gene takes into account the PCR efficacy, calculated based on dilution curves. The gene expression levels were normalised towards a reference gene. For all dietary treatments, three different reference genes were measured; acidic ribosomal protein (ARP), β-actin and elongation factor 1-αβ isoform (EF1AB) were the reference genes that were analysed based on previous Atlantic salmon reference gene validation(Reference Olsvik, Lie and Jordal50). The reference gene stability was tested by geNorm (Reference Vandesompele, Preter and Pattyn51), and all reference genes were found to have stability within acceptable limits. β-Actin, however, exhibited low PCR efficacy in liver and was hence disqualified as a reference gene. The most stable of these three reference genes (EF1AB) was used for normalisation with the Q-Gene software (http://www.qgene.org).
All statistical analyses were performed using the program Statistica (version 9.0; Statsoft Inc., Tulsa, OK, USA). Significant differences among dietary treatments were assessed by a one-way ANOVA(Reference Zar52). Where the null hypothesis (H0, no difference between treatments or within treatment at different time intervals) was rejected, significant differences were tested using Tukey's honestly significant difference test (P < 0·05; Sokal & Rohlf(Reference Sokal and Rohlf53)). A Kolmogorov–Smirnov test was used to assess the normality of distribution of each treatment(Reference Zar52). All data were normally distributed. Dependent variables were checked for homogeneity of variance by the Levene test and transformed whenever necessary(Reference Zar52).
Results
After 3 months of feeding, the pellet size was adjusted according to fish size, from 4 mm to 6 mm pellets. Lipid level increased as pellet size increased, resulting in a slightly higher lipid:protein ratio, while the ratio of plant ingredients relative to marine ingredients was kept constant. Dietary amino acids differed due to the substitution of plant protein for fish meal and reflected the amino acid composition of the different protein ingredients in the various experimental diets. The indispensable amino acids:dispensable amino acids ratio was lower in the replacement groups than in the FMFO group(Reference Torstensen, Espe and Sanden36). The higher the inclusion level of plant protein, the lower the amount of taurine in these diets and the lower the concentration of non-amino acid-N. However, only minor differences in indispensable amino acids were observed between the 4 and 6 mm feeds(Reference Torstensen, Espe and Sanden36). The vegetable oil blend was formulated to mimic fish oil in total SFA, MUFA and PUFA content but with no highly unsaturated n-3 PUFA, and this was largely achieved. Replacement of fish oil with the vegetable oil blend resulted in increased percentages of 18 : 3n-3, 18 : 2n-6 and 18 : 1n-9, with concomitant decreased proportions of highly unsaturated n-3 PUFA and long-chain monoenoic fatty acids such as 20 : 1 and 22 : 1. These differences were quantitatively greater in the diets with the higher level of fish oil replacement, in diets 40PP70VO and 80PP70VO. In contrast to the replacement of protein sources, substitution of fish oil with the vegetable oil blend had similar effects in the 6 and 4 mm diets. Total levels of monoenoic fatty acids, however, were similar in the 4 mm diets and elevated in the 6 mm plant diets, mainly due to 18 : 1 n-9 (Table 3). The ash content of diets was highest in the FMFO diet and decreased with increasing plant protein inclusion (Table 1). Similarly, non-starch, ash, protein or fat DM increased when fish meal was replaced by plant protein and krill meal (Table 1).
Table 3 Fatty acid composition (area % wet weight), total fatty acids (mg/g) and amino acids (g/16 g nitrogen) of the four experimental diets at two different pellet sizes
FMFO, 100 % fish meal and 100 % fish oil; FMFO, 100 % fish meal and 100 % fish oil; 80PP35VO, 80 % plant protein and 35 % vegetable oil blend; 40PP70VO, 40 % plant protein and 70 % vegetable oil blend; 80PP70VO, 80 % plant protein and 70 % vegetable oil blend.
* Lysine, methionine and taurine (g/16 g N) contents of the experimental diets are listed. The rest of the dietary amino acid compositions are given in Torstensen et al. (Reference Torstensen, Espe and Sanden36).
Dietary lipid class composition was significantly affected by the raw materials used (Table 4). All phospholipids decreased with decreasing fish meal and fish oil inclusion whereas NEFA concentration increased. Total amount of sterols (cholesterol+phytosterols) in all experimental diets remained the same. The phytosterol composition of the diets is reported elsewhere (BE Torstensen, M Espe and Ø Lie, unpublished results). When quantifying lipid classes by high-performance TLC, the amount of a lipid class is dependent on the number of double bonds in the lipid class. Assuming that the difference in dietary fatty acids (Table 3) is representative also for the phospholipids, approximately 20 % of the reduction in phosphatidylcholine may be due to a reduced number of double bonds.
Table 4 Dietary phospholipid and sterol composition (mg/g) of the four experimental diets
FMFO, 100 % fish meal and 100 % fish oil; 80PP35VO, 80 % plant protein and 35 % vegetable oil blend; 40PP70VO, 40 % plant protein and 70 % vegetable oil blend; 80PP70VO, 80 % plant protein and 70 % vegetable oil blend; SM, sphingomyelin; nd, not detected; PC, phosphatidylcholine; PI, phosphatidylinositol; PS, phosphatidylserine; PE, phosphatidylethanolamine; STEROLS, cholesterol and phytosterol.
Whole-body lipid stores – adiposity
High inclusion levels of plant proteins combined with the vegetable oil blend (80PP70VO) resulted in significantly increased whole-body lipids compared with high inclusion of plant proteins together with low vegetable oil (80PP35VO) (Fig. 1(A)). This was not reflected in total fillet lipid level (Fig. 1(B)). Indeed, the visceral somatic index was significantly higher in fish fed 80PP70VO compared with the other dietary groups (Fig. 1(C)). It is well known that fillet and whole-fish lipid level can be positively correlated with fish weight. Fish fed 80PP70VO and 80PP35VO had significantly lower final weight(Reference Torstensen, Espe and Sanden36) and may therefore have decreased relative fillet and whole-fish lipid level due to this. Therefore all lipid data were presented per kg fish to eliminate this confounding factor.
Fig. 1 Whole-fish lipid level (A), fillet lipid level (B) and visceral somatic index (VSI) (C), all corrected for differences in final fish weight, from Atlantic salmon (Salmo salar L.) fed either 100 % fish meal and 100 % fish oil (FMFO), 80 % plant proteins and 35 % vegetable oil blend (80PP35VO), 40 % plant proteins and 70 % vegetable oil blend (40PP70VO) or 80 % plant proteins and 70 % vegetable oil blend (80PP70VO) for 12 months. Values are means (n 3), with standard deviations represented by vertical bars. a,b,c Mean values with unlike letters were significantly different (P < 0·05; one-way ANOVA).
Liver lipid stores and fatty acid composition
Liver lipid stores (TAG) increased in fish fed the plant ingredient diets differently through the 12 months, with TAG being more than twofold in the 80PP70VO group compared with the FMFO control after 12 months of feeding (Fig. 2). It is important to note that TAG accumulation was affected differently during the seawater period (Fig. 2). After 3 months, all groups of fish fed diets containing plant protein and vegetable oils showed significantly increased TAG concentration compared with the FMFO-fed fish. After 8 months, however, no significant differences were observed but there were large standard deviations in the 80PP70VO group. After 12 months, the combined high-plant protein and high-vegetable oil group (80PP70VO) showed significant TAG accumulation compared with the FMFO and 80PP35VO groups. Hepatic TAG level of the 40PP70VO-fed fish, however, was in between the other dietary treatments and was not statistically different. Considering that differences in liver fatty acid composition (Table 5) influence the quantification of TAG, TAG in the two 70VO groups would be underestimated by approximately 14 % (due to 14 % fewer double bonds in the livers of 70VO compared with fish oil-fed fish). In this case, the consequence would be even larger differences in TAG level between fish oil- and 35 % vegetable oil (35VO)-fed fish compared with the two groups of 70VO-fed fish.
Fig. 2 Liver lipid stores (TAG) from Atlantic salmon (Salmo salar L.) fed either 100 % fish meal and 100 % fish oil (□), 80 % plant proteins and 35 % vegetable oil blend (
), 40 % plant proteins and 70 % vegetable oil blend (
) or 80 % plant proteins and 70 % vegetable oil blend (■) for 12 months. Values are means (n 3), with standard deviations represented by vertical bars. a,b Mean values with unlike letters were significantly different (P < 0·05; one-way ANOVA).
Table 5 Liver fatty acid composition (area %) and nitrogen metabolites (μmol/100 g wet weight) from Atlantic salmon (Salmo salar L.) after being fed the four experimental diets for 12 months
(Mean values and standard deviations for three fish per diet)
FMFO, 100 % fish meal and 100 % fish oil; 80PP35VO, 80 % plant protein and 35 % vegetable oil blend; 40PP70VO, 40 % plant protein and 70 % vegetable oil blend; 80PP70VO, 80 % plant protein and 70 % vegetable oil blend.
a,b,c,d Mean values within a row with unlike superscript letters were significantly different (P < 0·05; one-way ANOVA).
Liver fatty acid composition was highly influenced by the dietary fatty acid composition after 3, 8 (data not shown) and 12 months of feeding (Table 5). Overall 18 : 1n-9, 18 : 2n-6 and 18 : 3n-3 increased. They were at similar levels in the two 70 % vegetable oil groups whereas the marine n-3 fatty acids and 20 : 4n-6 were significantly decreased when vegetable oils replaced fish oil (Table 5). In the FMFO diets, 20 : 4n-6 was higher (Table 3), whereas the levels of 18 : 2n-6 were 6-fold higher in the 70 % vegetable oil diets compared with the fish oil diet and 6-fold higher in the salmon livers. The concentration of 18 : 3n-3, however, was almost 9-fold higher in the vegetable oil diets compared with fish oil and 6-fold higher in the salmon livers (Table 5). The 18 : 1n-9 content was approximately 3-fold higher in the 70 % vegetable oil diets, whereas 2-fold and significantly higher in the livers of salmon fed the 70VO diets compared with FMFO controls (Tables 3 and 5).
Liver nitrogen metabolites
Liver taurine concentration was significantly lower in fish fed the two diets with high plant protein inclusion (diets 80PP35VO and 80PP70VO) at the final sampling after 12 months of feeding. The non-protein-bound glycine, serine, methionine and lysine were not different in fish fed any of the plant diets (Table 5). Liver cystathionine concentration was significantly lower in fish fed the high-plant protein, low-vegetable oil diet (80PP35VO) as compared with the fish fed the total replacement diet (80PP70VO). No significant differences in liver phosphatidylethanolamine concentration were detected in any of the dietary groups.
Plasma and VLDL TAG
Plasma and VLDL TAG significantly increased when both fish meal and fish oil were replaced with maximum levels of vegetable oil (70 %) and 80 % plant protein mix (Fig. 3), being significantly higher after 3 and 12 months of feeding. The two intermediate replacement groups (80PP35VO and 40PP70VO) varied more in plasma and VLDL TAG concentration and with higher levels after 8 months compared with 12 months of feeding. There were no significant differences between 80PP35VO and 40PP70VO and FMFO or 80PP70VO at any of the three sampling periods (Fig. 3). The sampling point after 8 months of feeding exhibited less difference between the groups and no statistical significant differences were observed (Fig. 3).
Fig. 3 Plasma TAG (mg/ml) (A) and VLDL TAG (mg TAG/g plasma) (B) after 3 months from Atlantic salmon (Salmo salar L.) fed either 100 % fish oil (FMFO; - -□- -) or 80 % plant proteins and 70 % vegetable oil blend (80PP70VO; - -♦- -), and after 8 and 12 months from Atlantic salmon fed either FMFO, 80 % plant proteins and 35 % vegetable oil blend (80PP35VO; –□–) or 40 % plant proteins and 70 % vegetable oil blend (40PP70VO; –■–) or 80PP70VO. Values are means (n 3), with standard deviations represented by vertical bars. Statistical differences within each time point were tested by one-way ANOVA. a,b Mean values with unlike letters were significantly different (P < 0·05). No letters were assigned for groups not being significantly different from the other groups (i.e. 80PP35VO and 40PP70VO).
Expression of genes encoding lipid transport and uptake
The increased plasma TAG in 80PP70VO compared with FMFO was opposite to the expression of the visceral adipose tissue fatty acid binding protein (FABP11), which was down-regulated in visceral adipose tissue of 80PP70VO-fed fish (Fig. 4(A)). The expression of FABP11 was approximately 10-fold higher compared with lipoprotein lipase (LPL) and fatty acid transport protein (FATP) in visceral adipose tissue (Fig. 4(B)). Furthermore, the expression of cd36 was only at trace levels compared with the other lipid uptake and transport proteins measured in visceral adipose tissue (Fig. 4(B)). No changes in the expression of fatty acid uptake genes (LPL, FATP1 and cd36; Fig. 4(B)) were present, which may indicate no difference in the uptake of fatty acids in visceral adipose tissue.
Fig. 4 Normalised gene expression levels (normalised against the reference gene elongation factor 1αB (EF1AB)) of fatty acid binding protein 11 (A) and of fatty acid binding protein 3 (FABP3), cd36, fatty acid transport protein 1 (FATP1) and lipoprotein lipase (LPL) (B) in visceral adipose tissue from Atlantic salmon (Salmo salar L.) fed either 100 % fish meal and 100 % fish oil (FMFO; ■), 80 % plant proteins and 35 % vegetable oil blend (80PP35VO; □), 40 % plant proteins and 70 % vegetable oil blend (40PP70VO; 
) or 80 % plant proteins and 70 % vegetable oil blend (80PP70VO; ) for 12 months. Values are means, with 95 % CI represented by vertical bars. a,b Mean values with unlike letters were significantly different (P < 0·05; one-way ANOVA).
In the liver, the expression of genes related to fatty acid uptake (cd36) and intracellular transport (FABP10) was not significantly affected when fish meal and fish oil were replaced with 80 % plant protein and 70 % vegetable oil (Fig. 5(A) and (B)); however, there seemed to be a trend of increased expression of cd36 and FABP10 in 80PP70VO-fed salmon. The level of expression of FATP1 and cd36 was comparable in liver, whereas FABP10 was expressed 100-fold higher. The expression of the gene encoding for apoB100, being a part of the VLDL particle, was significantly up-regulated in the 80PP70VO-fed fish compared with the FMFO-fed fish, while the intermediate groups (80PP35VO and 40PP70VO) were in between (Fig. 5(C)). The expression of LPL in the liver was not affected by dietary plant raw materials (data not shown).
Fig. 5 Normalised gene expression levels (normalised against the reference gene elongation factor 1αB (EF1AB)) of fatty acid transport protein 1 (■) and cd36 () (A), fatty acid binding protein 10 (B) and apoB100 (C) in the liver from Atlantic salmon (Salmo salar L.) fed either 100 % fish meal and 100 % fish oil (FMFO), 80 % plant proteins and 35 % vegetable oil blend (80PP35VO), 40 % plant proteins and 70 % vegetable oil blend (40PP70VO) or 80 % plant proteins and 70 % vegetable oil blend (80PP70VO) for 12 months. Values are means, with 95 % CI represented by vertical bars. a,b Mean values with unlike letters were significantly different (P < 0·05; one-way ANOVA).
Discussion
Replacing fish meal and fish oil with high levels of a plant protein mixture and vegetable oil blend resulted in increased overall adiposity in Atlantic salmon post-smolt including increased visceral adipose tissue, liver lipids and plasma TAG contents after 12 months of feeding. Intermediate replacement levels (40PP70VO and 80PP35VO) did not have the same effects, indicating that Atlantic salmon metabolism was affected in such a way that the fish developed increased adiposity through interactions between high levels of plant protein and vegetable oil inclusion. To our knowledge, no previous studies have been performed with combined high replacement of fish meal and fish oil in Atlantic salmon focusing on fish lipid metabolism and deposition following a long-term feeding experiment. Thus, the present study has demonstrated for the first time the interaction between plant proteins and vegetable oils in the lipid metabolism of fish. In mammalian models, reports show that either protein or lipid source affects lipid deposition and metabolism(Reference Gudbrandsen, Wergedahl and Mork54–Reference Tovar and Torres58), but even in these models the combined effect was very seldom studied(Reference Wergedahl, Gudbrandsen and Røst59).
Atlantic salmon fed high levels of both plant protein and vegetable oil seem to store significantly more lipids in visceral adipose tissue. In rodents, fish oil is known to decrease adiposity through reducing the adipocyte cell size and thereby improving insulin sensitivity and decreasing the release of NEFA(Reference Lombardo, Hein and Chicco60). Atlantic salmon adipocytes decrease their lipid accumulation when stimulated with marine n-3 fatty acids compared with 18 : 1n-9 in vitro (Reference Todorcević, Vegusdal and Gjøen61) and this negatively correlated with FATP1 expression, indicating a role of FATP1 in adipocyte lipid accumulation(Reference Huang, Todorcevic and Ruyter62). The suppression of TAG accumulation by n-3 PUFA has also been reported in studies with 3T3-L1 pre-adipocytes(Reference Kim, La-Fera and Lin63), and in mammalian studies where n-3 PUFA limits the hypertrophy of fat depots compared with high-fat diets containing SFA(Reference Belzung, Raclot and Groscolas64). Hence, in several rodent models n-3 PUFA decreased adipocyte differentiation(Reference Okuno, Kajiwara and Imai65) and reduced fat accumulation(Reference Carlotti, Hainault and Guichard66–Reference Jones69). However, although differences in dietary n-3 PUFA levels may be an important factor in the development of visceral adipose lipid accumulation also in Atlantic salmon, dietary fatty acids alone cannot explain the lipid accumulation in the 80PP70VO group since 40PP70VO had the same dietary fatty acid composition and fish oil level but not the same visceral lipid accumulation. The present increased visceral lipid levels in 80PP70VO-fed salmon indicate that although known requirement levels were met in the current diets, combined decreased levels of fish oil together with low fish meal levels induce metabolic changes resembling lysine deficiency(Reference Espe, Lemme and Petri30) and/or n-3 PUFA deficiency at dietary levels considered adequate based on current knowledge(39). Thus, adequate dietary levels of one critical nutrient (for example, methionine and lysine or taurine or EPA and DHA) to avoid the development of fatty liver and or reduce visceral adiposity may be underestimated when other critical nutrients (for example, methionine or EPA and DHA) are available. This synergistic effect when high levels of plant proteins and vegetable oils replace fish meal and fish oil clearly shows the importance for nutrient requirement studies taking nutrient interaction effects into account when marine ingredients are replaced.
Visceral fat levels increased, whereas the expression of the gene encoding for FABP11 decreased in visceral adipose tissue when salmon were fed 80PP70VO. Down-regulation of FABP11 has also previously been reported in visceral adipose tissue and myosepta of Atlantic salmon fed a 100 % vegetable oil blend(Reference Torstensen, Nanton and Olsvik11). In mammals it is well established that the adipocyte FABP (AFABP/aP2) forms a physical complex with hormone-sensitive lipase that affects basal and hormone-stimulated adipocyte fatty acid efflux(Reference Smith, Sanders and Juhlmann70, Reference Baar, Dingfelder and Smith71) and specifically that knock-down of AFABP in mice resulted in decreased adipocyte fatty acid efflux(Reference Baar, Dingfelder and Smith71). Although not yet studied in Atlantic salmon, the adipocyte-specific FABP in Atlantic salmon (FABP11)(Reference Torstensen, Nanton and Olsvik11, Reference Agulleiro, André and Morais72) may have a similar function of regulating the adipocyte efflux of NEFA. The down-regulation of FABP11 expression, but not of the genes encoding for fatty acid uptake (FATP1, LPL, cd36), indicate that the mechanism of increased lipid level in the visceral adipose tissue of 80PP70VO-fed salmon was mainly decreased efflux of lipids from the adipose tissue rather than increased uptake.
Atlantic salmon fed high levels of both plant protein and vegetable oil had significantly increased plasma, VLDL and liver TAG levels. In contrast, completely replacing fish oil with vegetable oils in diets with high dietary fish meal inclusions was reported to have no effect on plasma and lipoprotein TAG concentration in Atlantic salmon, but significantly increased liver TAG(Reference Jordal, Lie and Torstensen9) and especially so at low water temperatures(Reference Ruyter, Moya-Falcón and Rosenlund10). This was confirmed in the 40PP70VO group of the present study, but only after 3 months of feeding. Importantly, the liver TAG stores developed during the 12 months' experiment, with all fish meal and fish oil replacement groups having significantly higher hepatic TAG levels after 3 months, whereas after 8 months no significant differences appeared due to high variation in the 80PP70VO group. After 12 months the liver TAG concentration from fish fed 80PP70VO was almost twofold higher than the other dietary groups, being statistically different from 80PP35VO- and FMFO-fed fish. Ruyter et al. (Reference Ruyter, Moya-Falcón and Rosenlund10) demonstrated the importance of water temperature for liver TAG storage when fish oil was replaced by soyabean oil. However, no increase in liver lipids was detected at 12°C but was present at 5°C. The present experiment was performed at a constant temperature at 9°C, which may explain the initial effects in the 80PP35VO and 40PP70VO groups that stabilised after 8 and 12 months of feeding. In the 80PP70VO-fed salmon, however, hepatic TAG accumulation was not reversed and was continuously increased also after 12 months of feeding at 9°C, indicating a synergistic effect of plant proteins and vegetable oil resulting in a metabolic imbalance which was independent of temperature.
Also, limitations in the indispensable amino acids lysine and methionine affect the lipid stores in post-smolt Atlantic salmon(Reference Espe, Lemme and Petri30, Reference Espe, Rathore and Du34, Reference Rathore, Liaset and Hevrøy35). Suboptimal dietary levels of methionine were reported to increase liver TAG concentration, fatty acid synthase activity and the 18 : 1:18 : 0 fatty acid ratio in Atlantic salmon, but had no impact on plasma TAG(Reference Espe, Rathore and Du34). Lysine limitation in post-smolt Atlantic salmon did not affect plasma or liver TAG or the fatty acid composition. Carcass lipid, however, increased upon lysine limitation(Reference Espe, Lemme and Petri30), whereas neither liver nor muscle fat increased(Reference Rathore, Liaset and Hevrøy35), indicating increased viscera adipose stores. Dietary levels of indispensible amino acids, including lysine and methionine, in the present study were similar in all diets and all were above defined requirements for on-growing Atlantic salmon(Reference Torstensen, Espe and Sanden36). Obviously, the arbitrary voluntary feed intake will affect the availability of nutrients for general metabolism. As reported by Torstensen et al. (Reference Torstensen, Espe and Sanden36), Atlantic salmon fed 80PP70VO had decreased voluntary feed intake and thus growth during the first 3 months of the experiment, resulting in significantly lower final body weight (about 17 % lower than FMFO-fed fish). However, the voluntary feed intake was equal in all treatment groups thereafter until the final sampling, indicating that it was not lower intake of indispensable amino acids that could explain the increased liver TAG at the end of the experiment. We have found that liver concentrations of neither free methionine(Reference Espe, Rathore and Du34) nor lysine(Reference Rathore, Liaset and Hevrøy35) were affected by limitations in these amino acids. However, postprandial plasma concentration after 5 h was significantly affected by both lysine and methionine limitations(Reference Espe, Rathore and Du34, Reference Rathore, Liaset and Hevrøy35). In the present study plasma free lysine was about half in the fish meal and fish oil replacement groups as compared with the FMFO-fed fish, but no differences between any of the replacement diets were present (FMFO, 410 μmol/l; 80PP35VO, 40PP70VO and 80PP70VO, range 210–240 μmol/l). On the other hand, only the fish fed the 80PP35VO diet had significantly less plasma free methionine (140 μmol/l) as compared with 80PP70VO- (250 μmol/l) and FMFO-fed fish (190 μmol/l). Therefore it is unlikely that deprivation of indispensable amino acids explains the fat accumulation in the liver of fish fed 80PP70VO. However, taurine concentration was lower in both diets containing the low level of fish meal inclusion (80PP). Although Atlantic salmon has the capacity to synthesise taurine through trans-sulfuration(Reference Espe, Hevrøy and Liaset73), liver taurine concentration was still significantly lower in fish fed these diets after 12 months of feeding. Cystathionine concentration in liver, the first metabolite following the rate-limiting enzyme in trans-sulfuration(Reference Mato, Corrales and Lu74), was also affected; however, it was only significantly decreased in fish fed 80PP35VO. This may indicate that trans-sulfuration was affected in fish fed the diets containing high fish meal replacement. Dietary and liver taurine was significantly decreased when fish meal was replaced by plant proteins. Taurine is conjugated to bile acids in fish and improves the export of bile from the liver, and as such is linked to both the digestibility and clearance of cholesterol(Reference Bogevik, Tocher and Langmyhr75–Reference Yokogoshi and Oda77). Deficiency of taurine has been reported to decrease the total bile acid excretion in faeces(Reference Chen, Matuda and Nishimura78). Increasing the bile acid concentration by dietary cholic acid in mice has been reported to decrease liver and plasma TAG concentrations(Reference Watanabe, Houten and Wang79). The intake of protein sources high in taurine has been linked to reduced visceral lipid accumulation in rodent models(Reference Liaset, Madsen and Hao55). However, taurine alone through reduced bile acid cannot explain the increased fat in 80PP70VO-fed fish, since 80PP35VO-fed salmon had significantly less whole-fish and visceral fat levels but the same dietary and liver taurine levels. Although methionine limitation reduced plasma and faecal bile acid concentrations in post-smolt Atlantic salmon, the reduction did not reach statistical difference(Reference Espe, Liaset and Hevrøy80). Also, enrichment with crystalline dl-methionine did not increase plasma bile acids significantly in post-smolt Atlantic salmon(Reference Espe, Rathore and Du34). In juvenile Atlantic salmon, on the other hand, addition of taurine reduced the whole-body lipid:protein ratio(Reference Espe, Ruohonen and El-Mowafi81). Since the two dietary groups where fish meal was replaced by 80PP had the same dietary amino acid levels but not the same response regarding whole-body and visceral fat levels, it is not likely that differences in bile acid concentration alone explain the increased visceral mass in the present study. Overall, these results indicate that at low EPA and DHA, combined with possible suboptimal levels of amino acids, the requirements of lysine and methionine increase compared with plant protein diets in which fish oil constitutes the lipid source(Reference Espe, Rathore and Du34). However, since no difference in the liver 18 : 1:18 : 0 ratio was observed, firm conclusions cannot be drawn, and further studies are required to elucidate the potential role of dietary amino acids on increased liver TAG in fish meal and fish oil replacement diets.
The three replacement diets contained about half the amount of phospholipids as compared with the FMFO control diet. The lower phosphatidylcholine may interact with protein metabolism because during low dietary choline or phosphatidylcholine intake the animals are able to synthesise phosphatidylcholine from phosphatidylethanolamine through three successive methylation reaction in the liver(Reference Noga and Vance82, Reference Vance83). In the present study liver phosphatidylethanolamine was similar in all groups, indicating similar phosphoethanolamine synthetase (pemt) activity and hence endogenous synthesis of phosphatidylcholine. However, in the present study neither phosphoethanolamine synthetase activity nor S-adenosylmethionine or choline status was analysed. On the other hand, we have previously reported that methionine availability affected liver S-adenosylmethionine(Reference Espe, Hevrøy and Liaset73) and reduced liver pemt activity (M Espe, EM Hevrøy and B Liaset, unpublished results). Furthermore, phospholipid deficiency and particularly choline deficiency are well-known causes of hepatic lipid accumulation and decreased plasma VLDL levels in rodents(Reference Anstee and Goldin84), especially when methionine is also low(Reference Chawla, Watson and Eastin85–Reference Slow and Garrow87). The phospholipid requirement has not been established for adult fish, but is considered to be essential for optimal growth and development at juvenile stages(Reference Tocher, Bendiksen and Campbell88). Thus, the increased TAG accumulation in 80PP70VO-fed salmon may be due to an interaction also with dietary phospholipids. However, a classical choline deficiency decreases plasma VLDL and hence plasma TAG, which was the opposite of the effect on plasma and VLDL TAG in 80PP70VO-fed fish. Hence, decreased dietary phospholipids may play a role together with other amino acids and fatty acids in the overall increase in liver TAG and VLDL TAG.
Increases in plasma and VLDL TAG levels were found in both groups of Atlantic salmon fed diets where 80 % of the fish meal protein was replaced by the plant protein mix; these increases were observed most profoundly and consistently in the 80PP70VO group. Several studies in human subjects have shown that dietary EPA and DHA decrease plasma TAG(Reference Harris, Connor and McMurphy24, Reference Nestel25) and protect against CHD(Reference Bang, Dyerberg and Nielsen26–Reference Seierstad, Seljeflot and Johansen27). More recently it was demonstrated that high dietary EPA and DHA decreased hepatic TAG secretion in Atlantic salmon compared with a standard fish oil diet and with a rapeseed oil-based diet(Reference Kjær, Vegusdal and Gjøen89). In mammals fish oil has been demonstrated to interfere with assembly of the initial VLDL precursor particles in hepatocytes(Reference Lang and Davis21, Reference Wang, Chen and Fisher90, Reference Wong, Fisher and Marsh91), all supporting that increased plasma TAG in the 80PP70VO-fed salmon may be a result of increased secretion of TAG-rich VLDL particles from the liver in the post-absorptive phase regulated by dietary fatty acid composition. The assembly process of hepatic VLDL in mammals is recognised to be initiated in the endoplasmic reticulum as soon as apoB100 is translated and translocated into the lumenal side where the elongating apoB100 polypeptide chain recruits various lipids co-translationally (for a review, see Sundaram & Yao(Reference Sundaram and Yao92)). Interestingly, the gene encoding for apoB100 was increasingly expressed when the inclusion of vegetable oil and plant proteins increased. Furthermore, although not statistically significant, there was a trend towards increased fatty acid uptake and intracellular transport of fatty acids through the increased RNA expression of cd36 and FABP10 in the livers of 80PP70VO-fed Atlantic salmon. In summary, this indicates a more active hepatic fatty acid turnover, accumulation and transport of neutral lipids in high-plant protein- and vegetable oil-fed fish compared with the traditional fish oil- and fish meal-fed fish as well as the lower replacement level diets. Previous studies replacing fish oil with vegetable oil but not replacing dietary fish meal have reported no changes in plasma or VLDL TAG(Reference Jordal, Lie and Torstensen9), confirming the findings of Kjær et al. (Reference Kjær, Vegusdal and Gjøen89) of no differences in hepatic TAG secretion when comparing fish oil and rapeseed oil-fed fish. In post-smolt Atlantic salmon, limitation in either lysine or methionine had no significant impact on plasma TAG(Reference Espe, Rathore and Du34, Reference Rathore, Liaset and Hevrøy35). In European seabass fed plant protein-based diets, plasma TAG was reduced as compared with fish meal-fed fish(Reference Dias, Alvarez and Arzel6). Again, significantly and consistently increased plasma and VLDL TAG levels in the combined high-plant protein- and vegetable-oil-fed fish demonstrate an interaction between dietary lipids and proteins with lipid metabolic consequences.
Intestinal lipid droplets have been reported to appear in vegetable oil-fed fish(Reference Ruyter, Moya-Falcón and Rosenlund10, Reference Caballero, Izquierdo and Kjørsvik93–Reference Olsen, Myklebust and Kaino95), which may be due to a slower transport of lipids from the intestinal cells to the circulation. Hence, the increased plasma TAG observed in 80PP70VO-fed salmon may be due to a slower lipid uptake detected 48 h after the last meal. Furthermore, the apparent digestibility of 16 : 0 was 16 % lower in 80PP70VO-fed fish compared with the other three groups(Reference Torstensen, Espe and Sanden36), possibly slightly decreasing the lipid uptake. Plasma TAG was also analysed 6 h after the last meal, revealing no significant differences (data not shown). This, together with the discrepancy in effect between the two high vegetable oil groups, indicates that increased plasma TAG was not solely due to a possible increased intestinal lipid accumulation in vegetable oil-fed fish.
In mammals, the metabolic syndrome is related to increased central (visceral) adiposity, increased plasma TAG, decreased HDL-cholesterol, hypertension, glucose intolerance and type 2 diabetes, which together are well-documented risk factors for CVD(Reference Cheal, Abbasi and Lamendola96). In conclusion, Atlantic salmon fed 80PP70VO demonstrated increased visceral adiposity, increased plasma and VLDL TAG as well as increased hepatic TAG stores and decreased plasma HDL-cholesterol levels (BE Torstensen, M Espe and Ø Lie, unpublished results). Compared with the metabolic syndrome in mammals, these changes in lipid metabolism in 80PP70VO-fed Atlantic salmon together indicate a metabolic imbalance that may affect fish health and especially cardiovascular health. These issues will be investigated in further studies.
Acknowledgements
The study was funded by the IP-EU project ‘AQUAMAX’ (016249-2). We would like to thank Arnor Gullanger (IMR, Matre Aquaculture Research Station) for excellent fish husbandry, and the technical assistance of Jacob Wessels (NIFES) is greatly appreciated. Members of the technical staff at NIFES are thanked for excellent assistance with the chemical analysis.
Authors' responsibilities were as follows: B. E. T., M. E., I. S. and Ø. L. planned the project and feeding experiment; B. E. T., M. S. and I. S. collected samples, B. E. T. and M. S. analysed the data and wrote the manuscript; Ø. L. was project coordinator and all authors contributed to the writing process.
There are no conflicts of interest to report.
